Introduction

The prevalence of obesity in U.S. adults increased from 12.8% in the early 1960s to 22.5% in the early 1990s (1) . Recent studies estimate that 33% of U.S. adults are now obese (2, 3) and that obesity-related illnesses (primarily type 2 diabetes, coronary artery disease [CAD], and hypertension) account for 5.5% to 7.8% of all U.S. health care expenditures (4, 5, 6) .

Until recently, most experts believed that obesity operated indirectly through other CAD risk factors, such as hypertension, dyslipidemia, and impaired glucose tolerance or non-insulin-dependent diabetes mellitus (7, 8) . However, longitudinal studies have demonstrated that obesity also has a direct relationship with CAD (9, 10, 11), and in 1997, the American Heart Association reclassified obesity as a major, modifiable CAD risk factor (12) . Despite extensive research on the role of obesity as a CAD risk factor, little longitudinal clinical research examining its importance in patients with established CAD has been conducted (9, 13, 14, 15) .

Previous obesity cost-of-illness studies have been either prevalence-based (rather than incidence-based) or have studied all obese individuals instead of focusing on higher-risk subgroups (e.g., those with CAD) (6); thus, these studies could not estimate the long-term obesity-related expenses in higher-risk populations who would potentially benefit most from weight-loss therapies. The purpose of this study was to compare the effects of increasing body mass index (BMI) on the long-term clinical and economic burden of illness in CAD patients who presented with an acute coronary syndrome and underwent initial diagnostic cardiac catheterization between 1986 and 1997.

Research Methods and Procedures

Patient Population

Our population included acute myocardial infarction (MI) (MI within the previous 6 weeks) and unstable angina (UA) patients with 75% stenosis in one or more major epicardial coronary segments (16, 17) who had an initial cardiac catheterization at Duke University Medical Center between January 1, 1986 and December 31, 1997. UA patients had an admission to rule out MI (with no MI documented in the medical record) or unstable chest pain. Acute MI patients had an evolving MI within the previous 24 hours, MI as the reason for admission, or an MI within the last 6 weeks. Exclusion criteria included BMI < 18.5 kg/m², a previous revascularization procedure (e.g., percutaneous coronary intervention [PCI] or coronary artery bypass graft [CABG] surgery), a diagnosis of congenital or valvular heart disease, idiopathic cardiomyopathy or pericardial disease, severe mitral regurgitation, or thrombolytic therapy within 90 days of cardiac catheterization.

We classified patients using the obesity categories of the National Institutes of Health (18) . In this system, patients with BMIs between 18.5 and 24.9 kg/m² are considered normal weight and patients with BMIs between 25.0 and 29.9 kg/m² are considered overweight. The National Institutes of Health classifies obese patients (BMI 30.0 kg/m²) into three categories (obese I, II, and III). We also classified patients with BMIs between 30.0 and 34.9 kg/m² as obese, but due to the small number of patients with BMIs > 35 kg/m², we collapsed the obesity II and III categories into one category, which we called very obese.

Clinical Data Collection

Baseline Clinical and Hospitalization Data.

Cardiology fellows prospectively collected baseline demographic, physical examination, medical history, 12-lead electrocardiogram data, and results from the initial cardiac catheterization (19, 20) . Resource utilization data, length of stay, admission and discharge dates, and other administrative data were obtained from the Duke Hospital Information System (DHIS).

Follow-Up Cardiac Event Data.

We collected follow-up information on death, MI, CABG, PCI, and other hospitalizations. All patients were contacted 6 months after their index event and then annually. Deaths and MIs were confirmed by an independent events committee. Follow-up data were >95% complete and consisted of between 1 and 12.9 years on study patients. Non-Duke rehospitalizations were documented in the Duke Information System for Clinical Computing (DISCC) through its annual patient follow-up procedures; Duke rehospitalizations were confirmed in DHIS (16, 17, 21) .

Data Analysis

Baseline characteristics are presented as percentages for discrete variables and as medians (25th and 75th percentiles) for continuous variables. Unadjusted survival estimates across strata were calculated using the Kaplan–Meier method with the log-rank test to assess differences. Adjusted survival estimates across strata were calculated using Cox proportional hazards models.

Economic Data Estimation

Inpatient Costs of Care.

Our analysis was limited to inpatient medical costs (physician fees and hospital costs). We developed a series of regression models to estimate these costs for both the index and follow-up hospitalizations (Duke and non-Duke) using cost-weight information from the Global Use of Strategies to Open Occluded Coronary Arteries (GUSTO) IIb Economics and Quality of Life (EQOL) Substudy (22) . GUSTO IIb compared heparin with recombinant hirudin for the treatment of acute coronary syndromes and between 1994 and 1996 enrolled 2250 U.S. patients of whom 1773 from 196 centers participated in the EQOL substudy. These patients were evenly matched in terms of age, gender, and risk factors to patients in this population. Hospitalization costs for the GUSTO IIb study population were calculated by converting cost–center charges from patient bills into costs using hospital-specific departmental cost-to-charge ratios. Inpatient physician fees were estimated using the North Carolina Medicare Fee Schedule. All costs were adjusted to 1997 values using the medical component of the U.S. Consumer Price Index.

We needed six models (Duke baseline hospital costs and physician fees, Duke follow-up hospital costs and physician fees, and non-Duke follow-up hospital costs and physician fees) to estimate patient-level inpatient medical costs because of differences in patient populations and data collection between DISCC, DHIS, and GUSTO IIb. Variables included in these models were inpatient cardiovascular events (death, nonfatal MI, CABG, PCI, and diagnostic catheterization), age, and gender. Because length of stay was only available for Duke cost-models, models that estimated costs at Duke had greater explanatory power than did non-Duke models. For comparison purposes, discounting, the reverse of compound interest, at the rate of 3% per annum was used to adjust cash-flows during the follow-up to their equivalent 1997 values (23) .

Censored Follow-Up.

The statistical objective was to estimate the long-term (10-year) average event rates and inpatient medical costs for patients in the four BMI groups. Because not all patients had 10 years of data due to staggered entry and censored follow-up, we used the partitioned estimator of Bang and Tsiatis (24) to calculate the mean and SE for both clinical events and cost data. We partitioned the follow-up interval into five discrete groups: 0 to 30 days, 31 days to 1 year, +1 year to 3 years, +3 years to 5 years, and +5 years to 10 years and excluded follow-up after 10 years. Typically, partitioning the follow-up interval decreases the variability of estimates (24) . Statistical tests among BMI groups were conducted using large sample approximations.

Subgroup and Sensitivity Analyses

We performed subgroup analyses to estimate the relationships between increasing age and gender and total inpatient medical costs in the four BMI groups. Within BMI groups, we stratified patients by gender and age group (<55, 55 to 64, 65 to 74, and 75 years of age) and estimated inpatient medical costs using the partitioned estimator. We also performed sensitivity analyses to assess whether the effect of BMI on acute (30-day) and short-term (1-year) inpatient medical costs and resource utilization was still evident after adjusting for year of catheterization. General linear models were constructed using BMI group and year of catheterization as discrete predictor variables.

Results

Baseline Characteristics and Acute Phase Outcomes

There were 9407 patients who met study entry criteria; 2 patients were excluded because of data quality issues. Of the 9405 patients, 2951 (31.4%) were classified as normal weight, 3950 (42.0%) as overweight, 1694 (18.0%) as obese, and 810 (8.6%) as very obese. The number of patients who were obese or very obese at their initial catheterization increased from 20% to 33% between 1986 and 1997.

In general, higher BMI at initial cardiac catheterization was associated with greater prevalence of cardiac risk factors but less severe CAD (Table 1). Obese patients tended to be younger and more often women or minorities than overweight or normal-weight patients. Hyperlipidemia, hypertension, and diabetes all increased with BMI, whereas other risk factors (e.g., cerebral and peripheral vascular disease) had minimal associations with BMI (Table 1) . Additionally, obese patients tended to have more single-vessel and less severe (three-vessel or left main) disease. Obese patients also had greater use of PCI procedures during the acute CAD phase with lower unadjusted baseline 30-day mortality rates (Table 2).

Table 1 - Baseline clinical characteristics by obesity classification.

Full table

Table 2 - Ten-year clinical event rates.

Full table

Ten-Year Survival

Differences in baseline BMI were associated with statistically significant differences in unadjusted 10-year survival (p = 0.001). Ten-year unadjusted survival for very obese patients (54%) was similar to normal-weight patients (55%), whereas the survival of obese patients (62%) was closer to that of the overweight patients (63%; Figure 1). However, these survival differences were no longer statistically significant after adjusting for differences in age, baseline risk factors (including cardiac history), extent of CAD, and year of catheterization (Figure 2).

Figure 1:.

Unadjusted 10-year survival curves, by body mass index (BMI) category.

Full figure and legend (79K)

Figure 2:.

Ten-year survival curves by body mass index (BMI) category, adjusted for risk factors and diagnostic cardiac catheterization results.

Full figure and legend (82K)

Clinical Events

During the first 30 days after patients' cardiac catheterization, the number of events was similar for patients regardless of BMI group, although the higher a patient's BMI, the higher the number of PCI procedures (from 0.43 for normal-weight patients to 0.56 for very obese patients; Table 2). Conversely, the use of CABG declined with increasing BMI (from 0.32 for normal-weight patients to 0.24 for very obese patients; Table 2). Thirty-day mortality was low for all patients, although slightly lower as BMI increased (0.04 for normal-weight patients vs. 0.03 for very obese patients).

For all patients, regardless of BMI, the majority of events occurred in the post-30-day period, when increasing BMI was associated with a greater number of nonfatal cardiac events (Table 2) . Although their 10-year survival rate was similar, normal-weight patients had an average of 2.33 clinical events during the post-30-day period, whereas very obese patients experienced an average of 3.37 events (Table 2) . These differences were evident for all nonfatal clinical events. Likewise, although overweight and obese patients had similar 10-year survival rates, overweight patients experienced an average of 2.68 post-30-day clinical events, whereas obese patients averaged 3.35 events. Again, the differences occurred in all nonfatal clinical events.

Cumulative 10-year event rates reflected differences in post-acute events, with increasing BMI associated with more total events (ranging from 3.24 for normal-weight to 4.32 for very obese patients). Approximately 69% of all hospitalizations in this population occurred at Duke University Medical Center vs. 31% at other hospitals. This percentage did not differ significantly among the four BMI groups.

10-Year Medical Costs

The trends in 10-year discounted and undiscounted medical costs were similar among BMI groups. In general, patients with higher BMIs had higher total 10-year discounted inpatient medical costs ($44,162 for normal-weight patients vs. $51,777 for very obese patients and $44,246 for overweight vs. $49,914 for obese patients; Table 3). Acute medical costs tended to decrease with increasing BMI, whereas post-acute costs tended to increase. For normal-weight and overweight patients, the discounted costs were higher in the acute vs. post-acute phase, whereas obese and very obese patients had higher post-acute phase costs. Cumulative 10-year medical costs (discounted and undiscounted) for obese and very obese patients were greater than those for normal-weight patients. Overweight and normal-weight patients had similar cumulative 10-year medical costs (discounted and undiscounted).

Table 3 - Ten-year total inpatient medical costs.

Full table

Sensitivity and Subgroup Analyses

Temporal Patterns.

The percentage of very obese individuals in our study population increased from 5.2% in 1986 to 12.5% in 1997. In sensitivity analysis, we found that increasing BMI was associated with greater use of PCI procedures and lower mortality during the acute CAD phase after adjusting for differences in year of catheterization (Table 4). We also found that the year of index catheterization and BMI group were significant predictors of inpatient medical costs, with acute and total 1-year medical costs tending to decrease (in constant dollars) over time, while post-acute (31 days through 1 year) costs remained relatively constant. Increasing BMI (especially among obese and very obese patients) was associated with lower 30-day medical costs; overweight and obesity were associated with lower 1-year medical costs. After adjusting for differences in age and gender, overweight and obesity were associated both with lower 30-day and total 1-year inpatient medical costs.

Table 4 - Acute phase resource use and mortality adjusted to 1997 values.

Full table

Gender and Age.

Study patients with elevated BMI were more often women and younger. In subgroup analyses, we found that 10-year inpatient medical costs for obese and very obese women were $9000 to $13,000 higher than those for normal-weight and overweight women. Although there was a similar trend in 10-year inpatient medical costs for men, the costs for obese and very obese men were only $3000 to $4000 higher than for normal-weight and overweight men. We also found that older age (>65 years) was associated with elevated 10-year medical costs for all BMI groups and that the difference in 10-year medical costs between normal-weight and very obese patients increased from $5000 in patients <55 years of age to $17,000 in patients >75 years.

Discussion

This study is the first to investigate the effects of increasing BMI on long-term clinical outcomes and the economic burden of illness in CAD patients; several of its findings are particularly noteworthy. First, differences in baseline BMI were associated with statistically significant differences in unadjusted 10-year survival (p = 0.001) with survival for the very obese approximating that of normal-weight patients and survival for obese patients nearer to that of overweight patients. However, after adjusting for differences in patient age, other cardiac risk factors, diagnostic catheterization results, and year of catheterization, there were no statistically significant differences in survival among our BMI groups. Thus, more prevalent cardiac risk factors in obese and very obese CAD patients at baseline seemed to counteract potential survival benefits derived from their younger age and less extensive CAD. Second, although 30-day clinical event rates were similar among BMI groups, individuals with higher BMIs had more clinical events after the initial 30-day period. Thus, their long-term clinical burden of illness is greater. Lastly, although cumulative 10-year medical costs were greater with increasing BMI, there were differences in the distribution of those costs among BMI groups as 30-day costs tended to decrease with increasing BMI whereas post-30-day costs increased.

CAD Survival

Several studies have investigated long-term survival in ACS patients but none have assessed the influence of elevated BMI in this cohort. Gruppo Italiono per lo Studio della Sorpavivenza nell'Infarto-1 Miocardico (GISSI-I) reported a 10-year survival rate for acute MI patients of 54% (25), and we previously reported a 10-year survival rate of 57% for acute MI and 60% for UA patients at Duke (26) . In the present study, we report 10-year unadjusted survival for normal-weight and very obese ACS patients similar to that of the GISSI-I and Duke MI cohorts but a greater 10-year unadjusted survival for overweight and obese patients.

Unlike MI patients with a history of diabetes, hypertension, or previous MI (27, 28, 29) whose 30-day mortality rate is higher than patients without those risk factors, patients with a history of smoking have lower 30-day mortality than nonsmokers (30) . One explanation for this is that smokers tend to be younger with higher left ventricular ejection fraction than nonsmokers, but after adjustment for differences in index clinical and catheterization variables, there were no differences in the 30-day survival of smokers and nonsmokers. In a similar vein, overweight, obese, and very obese patients in our study had lower 30-day mortality rates compared with normal-weight patients; however, these patients were also younger and had less extensive CAD than normal-weight patients. Although there has been extensive research regarding the contribution of smoking to the pathogenesis of CAD and sudden cardiac death (through promotion of atherosclerosis, triggering of various cardiac events, and reduction in the capacity of blood to deliver oxygen) (31), the contribution of obesity is less understood. Blood volume and cardiac output increase with BMI (32), and obesity is known to affect diastolic function and to be associated with left ventricular hypertrophy (32) (particularly when BMI is >30 kg/m²) and a prolonged QT-interval (33) . Although recent research has found an independent association between elevated BMI (>25 kg/m²) and impaired coronary endothelial function, additional research is clearly needed to explain the specific pathophysiology behind the more aggressive form of the disease that becomes clinically evident at an earlier stage in the patients with higher BMIs in this study (34) .

Cost of Illness

Previous obesity cost-of-illness studies have estimated this disease's annual burden to society at 5.5% to 7.8% of total U.S. healthcare expenditures (6) . However, these have typically been broad, population-based studies that either used group- vs. patient-level data or were limited to costs in a single year. Quesenberry and colleagues (35) used information from a health maintenance organization in a single year to assess the relationship between BMI and costs. These researchers found that compared with normal-weight patients, obese patients had 33% greater and the very obese had 70% greater annual inpatient costs. In our CAD population, post-30-day inpatient medical costs were 14% higher for overweight, 53% higher for obese, and 63% higher for very obese patients.

Thompson et al. (15) developed a model to estimate the lifetime economic burden of obesity using information from National Health and Nutrition Examination Study (NHANES) III and the Framingham Heart Study. Their estimated lifetime discounted medical costs for patients 55 to 64 years old ranged from $21,900 to $39,000 by BMI group and gender. In our CAD population, 10-year discounted inpatient medical costs ranged from $44,162 to $51,777 by BMI group. Thompson et al. (36) also assessed the relationship between BMI and cumulative 9-year healthcare costs in a population 35 to 64 years old without a history of CAD and found that annual costs were $1641 in normal-weight patients but that costs for overweight patients were 10% higher and for obese patients and very obese patients were 36% higher. Because this population excluded patients with known CAD, their cost estimates were much less than in our CAD population. Nonetheless, both studies report similar increases in the chronic medical costs for overweight and obese patients compared with those for normal-weight patients.

Because our study used longitudinal patient-level data for obese patients with CAD, we were able to examine relationships between the level of obesity and long-term clinical event rates and health care expenditures more precisely than could be accomplished by previous methods. In doing this, we have provided an economic baseline against which to assess new therapeutic interventions (6, 20) .

Limitations

Our study is limited in that it is based on the experience of Duke University Medical Center between 1986 and 1998 and may not accurately reflect the experience of other centers or what will occur in the future in a rapidly evolving healthcare environment. Nonetheless, our results are grounded in a single, longitudinal clinical database with cost-weights derived from the shared experience of 196 medical centers. Second, we only report baseline BMI and not changes in BMI that might have occurred throughout the follow-up period and influenced clinical and economic outcomes. Nonetheless, we have no reason to believe that the experience of our patients will differ in any significant way from the typical ACS cohort with regard to changes in weight. Third, our study only included inpatient medical costs and does not include costs associated with outpatient physician visits and medications. Although outpatient costs are less important in CAD patients, their inclusion would most likely have increased the total medical costs associated with increasing BMI. Fourth, our economic data are modeled and do not measure the costs associated with lost productivity (through morbidity or early mortality). However, all inpatient costs were taken from the GUSTO IIb clinical trial and were fitted to the specific exigencies of Duke's clinical practice and, therefore, should provide reasonable estimates of this experience. Finally, because the costs associated with lost productivity have been directly associated with obesity in previous studies, we decided to concentrate our efforts where new information was needed and where our data could provide results.

In conclusion, in this first examination of the clinical and economic burden of obesity in ACS patients, we found that the burden of elevated BMI in CAD continues long after the initial acute event has resolved. Whether, in CAD patients, therapies directed at the obesity state itself, as contrasted with concomitant conditions such as hypertension or diabetes mellitus, can reduce the morbid and economic consequences of obesity deserves empirical testing. Our study suggests that the potential for an effective therapy may be substantial.

References

1. American Heart Association (2001) 2001 Heart and Stroke Statistical Update. American Heart Association: Dallas, TX.
2. Kuczmarski, R. J., Flegal, K. M., Campbell, S. M., Johnson, C. L. (1994) Increasing prevalence of overweight among US adults: the National Health and Nutrition Examination Surveys, 1960 to 1991. J Am Med Assoc. 272: 205–211. | Article | ISI | ChemPort |
3. Rosenbaum, M., Leibel, R. L., Hirsch, J. (1997) Medical progress: obesity. N Engl J Med. 337: 396–407. | Article | PubMed | ISI | ChemPort |
4. Wolf, A. M. (1998) What is the economic case for treating obesity? Obes Res. 6: 2S–7S. | PubMed |
5. Wolf, A. M., Corditz, G. A. (1998) Current estimates of the economic cost of obesity in the United States. Obes Res. 6: 97–106. | PubMed | ISI | ChemPort |
6. Kortt, M. A., Langley, P. C., Cox, E. R. (1998) A review of cost-of-illness studies in obesity. Clin Ther. 20: 772–779.
7. Reavin, G. M. (1988) Banting lecture: role of insulin resistance in human disease. Diabetes. 37: 1595–1607. | Article | PubMed | ISI | ChemPort |
8. Lew, E. A., Garfinkel, L. (1979) Variations in mortality by weight among 750,000 men and women. J Chron Dis. 32: 563–576. | Article | PubMed | ISI | ChemPort |
9. Garrison, R. J., Castelli, W. P. (1985) Weight and 30-year mortality of men in the Framingham Study. Ann Intern Med. 103: 1006–1009. | PubMed | ChemPort |
10. Manson, J. E., Willett, W. C., Stampfer, M. J., et al (1995) Body weight and mortality among women. N Engl J Med. 333: 677–685. | Article | PubMed | ISI | ChemPort |
11. Rabkin, M. T., Mathewson, F. A., Hsu, P. H. (1977) Relation of body weight to development of ischemic heart disease in a cohort of young North American men after a 26 year observation period: the Manitoba Study. Am J Cardiol. 39: 452–458. | Article | PubMed | ChemPort |
12. Eckel, R. H., Krauss, R. M. (1998) American Heart Association call to action: obesity as a major risk factor for coronary artery disease. Circulation. 97: 2099–2100. | PubMed | ISI | ChemPort |
13. Pi-Sunyer, F. X. (1993) Medical hazards of obesity. Ann Intern Med. 119: 655–660. | PubMed | ChemPort |
14. Feinleib, M. (1985) Epidemiology of obesity in relation to health hazards. Ann Intern Med. 103: 1019–1024. | PubMed |
15. Thompson, D., Edelsberg, J., Colditz, G. A., Bird, A. P., Oster, G. (1999) Lifetime health and economic consequences of obesity. Arch Intern Med. 159: 2177–2183. | Article | PubMed | ChemPort |
16. Harris, P. J., Harrell, F. E., Jr, Lee, K. L., Behar, V. S., Rosati, R. A. (1979) Survival in medically treated coronary artery disease. Circulation. 60: 1259–1269. | PubMed | ChemPort |
17. Califf, R. M., Harrell, F. E., Jr, Lee, K. L., et al (1989) The evolution of medical and surgical therapy for coronary artery disease: a 15-year perspective. J Am Med Assoc. 261: 2077–2086.
18. National Heart, Lung, and Blood Institute Obesity Education Initiative Expert Panel (1998) Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults: The Evidence Report. National Heart, Lung, and Blood Institute: Rockville, MD.
19. Rosati, R. A., McNeer, J. F., Starmer, C. F., Mittler, B. S., Morris, J. J., Jr, Wallace, A. G. (1975) A new information system for medical practice. Arch Intern Med. 135: 1017–1024.
20. Fortin, D. F., Califf, R. M., Pryor, D. B., Mark, D. B. (1995) The way of the future redux. Am J Cardiol. 76: 1177–1182.
21. Mark, D. B., Nelson, C. L., Califf, R. M., et al (1994) The continuing evolution of therapy for coronary artery disease: initial results from the era of coronary angioplasty. Circulation. 89: 2015–2025. | PubMed | ISI | ChemPort |
22. Mark, D. B., Granger, C. B., Ellis, S. G., et al (1996) Costs of direct angioplasty versus thrombolysis for acute myocardial infarction: results from the GUSTO II randomized trial. Circulation. 94: 168(abstr).
23. Gold, M. R., Siegel, J. E., Russell, L. B., Weinstein, M. C. (1996) Cost-Effectiveness in Health and Medicine. Oxford University Press: New York, NY.
24. Bang, H., Tsiatis, A. A. (2000) Estimating medical costs with censored data. Biometrika. 87: 329–343.
25. Franzosi, M. G., Santoro, E., De Vita, C., et al (1998) Ten-year follow-up of the first megatrial testing thrombolytic therapy in patients with acute myocardial infarction: results of the Gruppo Italiano per lo Studio della Sopravvivenza hell'Infarto Miocardico (GISSI-I) study. Circulation. 98: 2659–2665.
26. Eisenstein, E. L., Shaw, L. K., Nelson, C. L., Hakim, Z., Hasselblad, V., Mark, D. B. (2001) Assessing the clinical and economic burden of coronary artery disease: 1986–1998. Med Care. 39: 824–835.
27. Barbash, G. I., Reiner, J., White, H. D., et al (1995) Evaluation of paradoxic beneficial effects of smoking in patients receiving thrombolytic therapy for acute myocardial infarction: mechanism of the "smoker's paradox" from the GUSTO-I trial, with angiographic insights. J Am Coll Cardiol. 26: 1222–1229.
28. Patel, V., Topol, E. J. (2001) Lessons from the GUSTO trials. In Topol, EJ (ed). Acute Coronary Syndromes. Marcel Dekker: New York, NY. pp. 725–764.
29. Gottlieb, S. H., Fallavollita, J., McDermott, M. F., Brown, M. L., Eberly, S., Moss, A. J. (1994) Cigarette smoking and age at onset of a first non-fatal myocardial infarction. Coron Artery Dis. 5: 687–694.
30. Andrikopoulos, G. K., Richter, D. J., Dilaveris, P. E., et al (2001) Inhospital mortality of habitual cigarette smokers after acute myocardial infarction: the "smoker's paradox" in a countrywide study. Eur Heart J. 22: 776–784.
31. Goldstein, M. G., Niaura, R. (1998) Smoking. In Topol, EJ (ed). Comprehensive Cardiovascular Medicine. Lippincott-Raven: Philadelphia, PA. pp. 159–184.
32. Stone, N. J. (1998) Diet, nutritional issues, and obesity. In Topol, EJ (ed). Comprehensive Cardiovascular Medicine. Lippincott-Raven: Philadelphia, PA. pp. 45–70.
33. Frank, S., Colliver, J. A., Frank, A. (1986) The electrocardiogram in obesity: statistical analysis of 1029 patients. J Am Coll Cardiol. 7: 295–299. | PubMed | ISI | ChemPort |
34. Suwaidi, J. A., Higano, S. T., Holmes, D., Jr, Lennon, R., Lerman, A. (2001) Obesity is independently associated with coronary endothelial dysfunction in patients with normal or mildly diseased coronary arteries. J Am Coll Cardiol. 37: 1523–1528. | Article | PubMed |
35. Quesenberry, C. P., Caan, B., Jacobson, A. (1998) Obesity, health service use, and health care costs among members of a health maintenance organization. Arch Intern Med. 159: 2177–2183.
36. Thompson, D., Brown, J. B., Nichols, G. A., Elmer, P. J., Oster, G. (2001) Body mass index and future healthcare costs: a retrospective cohort study. Obes Res. 9: 210–218. | PubMed | ChemPort |

Acknowledgments

This work was supported in part by grants from Roche Global Pharmacoeconomic Research, Palo Alto, CA and Roche Laboratories, Nutley, NJ. This work was presented in part at the American Heart Association Scientific Sessions, November 1999, Atlanta, GA. We thank Tracey Dryden for assistance in drafting and editing the manuscript.